RE: live-boot todo for wheezy
I added the cc back to the list, in case anyone else has their thoughts.
> -----Original Message-----
> From: intrigeri [mailto:firstname.lastname@example.org]
> Sent: Friday, June 15, 2012 4:11 PM
> To: Ian Geiser
> Cc: intrigeri
> Subject: Re: live-boot todo for wheezy
> Ian Geiser wrote (15 Jun 2012 17:59:43 GMT) :
> >> > 2) Take snapshot of current running system.
> > Our product has 3 modes of operation for the client. The first one is
> > no persistence where we run on top of a tmpfs overlay. The other mode
> > is full persistence where we store all changes on the live-rw.
> > In the last mode we do the snapshots on demand. So the customer can
> > configure the system and then lock in that configuration. This way
> > when they reboot they go back to the snapshot state. What is
> > here is that we allow the customer to take as many snapshots as they
> > want.
> OK. I don't think this usecase is supported by custom mounts.
> You just volunteered to update and maintain the snapshots feature in
> live-boot, didn't you? :)
*sigh* would not be the first time I have volunteered myself for this sort of thing :)
I guess the big part is it sounds like snapshotting is not in the vision of the current persistence. What would happen if we just removed the current incarnation, and redid it with the new infrastructure in place. The implementation I have in my head would be similar to what we have now. A live-boot cpio command that unpacks the snapshot into memory after the ram overlay is setup. Then a userspace script that can on demand copy selective parts of the overlay into the cpio file. The selected parts can be configured through the new persistence configuration. Am I missing something? I am sure I am ;)